Harris AHS, Kuo AC, Bowe TR, et al. - In this study, the accuracy and parsimony of several machine learning strategies were compared for developing predictive models of failing to experience minimal clinically important differences in patient-reported outcome measures (PROMs) 1 year after total knee arthroplasty (TKA). Researchers included a total of 587 individuals in 3 large Veteran Health Administration facilities completing PROMs before and 1 year after TKA (92% follow-up). They applied several machine learning strategies for model development. They further used ten-fold cross-validation and bootstrapping to produce measures of overall accuracy (C-statistic, Brier Score). Furthermore, the sensitivity and specificity of various predicted probability cut-points were explored. It was reported that after TKA, models produced in this project provide estimates of patient-specific improvements in major outcomes 1 year. The data considered that integrating these models into clinical decision support, informed consent, and shared decision making could improve patient selection, education, and satisfaction.